CN106789764B - Joint Weighted Threshold denoises and the transform domain quadratic estimate method of balanced judgement - Google Patents
Joint Weighted Threshold denoises and the transform domain quadratic estimate method of balanced judgement Download PDFInfo
- Publication number
- CN106789764B CN106789764B CN201611031299.2A CN201611031299A CN106789764B CN 106789764 B CN106789764 B CN 106789764B CN 201611031299 A CN201611031299 A CN 201611031299A CN 106789764 B CN106789764 B CN 106789764B
- Authority
- CN
- China
- Prior art keywords
- sequence
- channel
- point
- estimation
- decision
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 101
- 230000004044 response Effects 0.000 claims abstract description 79
- 238000012549 training Methods 0.000 claims abstract description 54
- 230000008569 process Effects 0.000 claims abstract description 22
- 238000001914 filtration Methods 0.000 claims abstract description 20
- 238000005070 sampling Methods 0.000 claims description 37
- 230000014509 gene expression Effects 0.000 claims description 28
- 239000011159 matrix material Substances 0.000 claims description 26
- 238000010586 diagram Methods 0.000 claims description 25
- 230000005540 biological transmission Effects 0.000 claims description 22
- 238000006243 chemical reaction Methods 0.000 claims description 19
- 230000009466 transformation Effects 0.000 claims description 19
- 238000012545 processing Methods 0.000 claims description 18
- 125000004122 cyclic group Chemical group 0.000 claims description 16
- 230000000737 periodic effect Effects 0.000 claims description 6
- 238000012935 Averaging Methods 0.000 claims description 4
- 238000013139 quantization Methods 0.000 claims description 2
- 230000002123 temporal effect Effects 0.000 claims description 2
- 230000017105 transposition Effects 0.000 claims 1
- 230000006870 function Effects 0.000 description 21
- 238000004891 communication Methods 0.000 description 18
- 238000005516 engineering process Methods 0.000 description 13
- 230000001427 coherent effect Effects 0.000 description 6
- 241001270131 Agaricus moelleri Species 0.000 description 4
- 238000001514 detection method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000001228 spectrum Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 2
- 239000000969 carrier Substances 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000008054 signal transmission Effects 0.000 description 2
- 101150110503 END3 gene Proteins 0.000 description 1
- 241000764238 Isis Species 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/0202—Channel estimation
- H04L25/024—Channel estimation channel estimation algorithms
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03178—Arrangements involving sequence estimation techniques
- H04L25/03248—Arrangements for operating in conjunction with other apparatus
- H04L25/03254—Operation with other circuitry for removing intersymbol interference
- H04L25/03267—Operation with other circuitry for removing intersymbol interference with decision feedback equalisers
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)
Abstract
The invention discloses the denoisings of joint Weighted Threshold and the transform domain quadratic estimate method of balanced judgement, carry out as follows: 1) combining overlying training sequence and harmonic sequence generates and sends sequence；2) in OFDM baseband system, by the LS channel estimation methods of overlying training sequence, the rough estimate of channel frequency response is obtained；3) channel frequency response rough estimate is counted as amplitude and phase compensation, then after widened degree is the window function of M, M point dct transform, and low-pass filtering, time domain denoising and sequence zero padding extension process is made to gained sequence, to the N point sequence after extensionMake N point idct transform, window and secondary amplitude and phase compensation is gone to obtain4) acquired results equilibrium is adjudicatedM point dct transform, low-pass filtering, time domain denoising, interpolation zero padding and N point idct transform are repeated, the final estimation of frequency domain response is obtainedThis invention removes data sequences and training sequence to influence on channel estimating performance, improves the performance of existing LS channel estimation, implementation complexity is lower.
Description
Technical Field
The invention belongs to the technical field of digital communication, and particularly relates to a transform domain secondary estimation method for combined weighted threshold denoising and equalization judgment, which is suitable for the fields of wireless and power line multi-carrier communication and the like.
Background
In recent years, wireless communication technology has been rapidly developed. Among them, the time-varying characteristic of the wireless channel is always a research hotspot of the technology, and is also a key for improving the capacity and reliability of the wireless communication system. Meanwhile, the propagation of signals in a wireless channel is a complex process: including signal scattering, reflection and diffraction, along with the surrounding environment and the mobile station's effects on signal transmission. Therefore, fast real-time channel estimation is one of the keys of wireless communication technology. In the wired communication, a low voltage power line carrier communication (LPLC) technology is a high-speed broadband communication method by using an existing widely distributed low voltage power line as a medium for transmitting signals. The information transmission is carried out by utilizing the existing low-voltage power line network, so that rewiring is not needed, and the cost is reduced. However, in the LPLC system, the channel environment is complex, the noise interference is serious, and an efficient channel estimation technique and method are urgently needed to ensure the performance of the entire LPLC system.
Orthogonal Frequency Division Multiplexing (OFDM) is a special multicarrier modulation technique, and its principle is to modulate high-speed serial transmission data to a plurality of mutually independent orthogonal subchannels for parallel transmission through serial-to-parallel conversion, and has the advantages of strong anti-interference capability, high spectrum utilization rate, fast transmission rate, and the like. In practical communication systems, OFDM systems are divided into coherent OFDM systems and non-coherent OFDM systems, for example, LPLC systems belong to coherent OFDM. With coherent OFDM systems, the state information (CSI) of the channel is known during demodulation, and the CSI can be obtained by channel estimation. The channel estimation is a process of estimating parameters of a channel by using data of a receiving end through various methods. The existing channel estimation methods mainly include a channel estimation method based on a pilot frequency/training sequence method, a channel estimation based on a decision feedback method, and a blind channel estimation. Aiming at two methods of channel estimation based on pilot frequency or training sequence, wherein the implementation complexity is low and the estimation performance is excellent based on the pilot frequency channel estimation, but because of the conditions of channel time variation, selective fading and the like, the pilot frequency needs to be continuously sent for quickly tracking the channel state, and the bandwidth and power loss is easily caused; the channel estimation method based on the superimposed training sequence superimposes the low-power training sequence and the data sequence at the transmitting end, does not occupy the rest frequency spectrum resources, improves the utilization rate of the system bandwidth, can quickly track the change of the channel state, but the training sequence and the data sequence can have certain influence on the channel estimation effect.
Disclosure of Invention
Aiming at the defects of the two existing channel estimation methods, the invention adopts a superimposed training sequence and Discrete Cosine Transform (DCT) transform domain channel estimation method, and introduces an equalization judgment method in the transform domain channel estimation to complete the transform domain estimation of secondary denoising. Because the superposed training sequence does not occupy bandwidth, and the data sequence is transformed by combining the training sequence and the harmonic sequence, the influence of the data sequence on channel estimation can be eliminated; the generation of high-frequency components can be effectively inhibited due to DCT/Inverse Discrete Cosine Transform (IDCT), and noise interference is further eliminated by setting a weighted noise threshold to judge a channel response amplitude threshold of a transmission sequence and introducing an equalization judgment method in the operation process of the IDCT. The method has the advantages of moderate complexity, high performance, good stability and the like. Therefore, the method can be used for application occasions of real-time tracking of channel state change, rapid signal detection and the like in wireless and power line communication transmission environments.
The invention improves the existing low-performance high-complexity channel estimation method, provides an improved transform domain channel estimation method, namely a transform domain secondary estimation method combining weighted threshold denoising and equalization judgment, and can be widely applied to occasions such as channel estimation, rapid signal detection and the like of wireless and power line communication.
The invention is realized by the following technical scheme:
a transform domain secondary estimation method combining weighted threshold denoising and equalization decision adopts a method of generating a sending sequence by jointly superposing a training sequence and a harmonic sequence, a channel frequency response rough estimation method based on time domain statistical mean Least Square (LS) estimation, and a transform domain secondary estimation method combining weighted threshold denoising and equalization decision, and is realized by the following steps:
1) jointly superposing the training sequence and the harmonic sequence to generate a sending sequence, and eliminating the influence of the data sequence and the training sequence on channel estimation;
2) in a general OFDM baseband system, the rough estimation of the channel frequency response is obtained by adopting the existing LS channel estimation method of the superposed training sequence;
3) performing amplitude phase compensation on the channel frequency response rough estimation, performing window function with the widening degree of M (M is a positive integer, M OFDM symbols are a frame), performing M-point DCT (discrete cosine transform), performing low-pass filtering, time domain denoising and zero-filling expansion on the obtained sequence, and performing N (N is a positive integer) point sequence after expansionPerforming N-point IDCT conversion, windowing processing and secondary amplitude and phase compensation
4) The obtained results are judged in an equalizing wayRepeating the M-point DCT transformation, the low-pass filtering, the time domain denoising, the interpolation zero-filling and the N-point IDCT transformation to obtain the final estimation of the frequency domain response
Preferably, the method for generating the transmission sequence by jointly superimposing the training sequence and the harmonic sequence obtains the data sequence { d (n) } after processing the binary input sequence { x (n) } through the front end of an Orthogonal Frequency Division Multiplexing (OFDM) baseband system. And superimposes the training sequence { t (n) } and the harmonic sequence { a (n) } jointly, resulting in the transmission sequence { s (n) }. Wherein, the harmonic sequence { a (n) } is a random periodic sequence, and the period is the same as the training sequence { t (n) }. The method can be realized by the following detailed steps:
step 2.1, let input sequence x (N) with length N (N is integer) be coded/interleaved, modulated, 1: N serial-to-parallel converted (A)1:A2Is represented by A1Way conversion to A2Road, and A1,A2The same applies hereinafter to natural numbers) and N-point Inverse Fast Fourier Transform (IFFT), to obtain a data sequence { d (N) }. The coding/interleaving can adopt the existing Turbo coding and the like and is matched with the interleaving modes of pseudo-random and the like; the modulation mode is selected from multi-system quadrature amplitude modulation (M-QAM) and the like; also, Turbo coding, pseudo random interleaving, M-QAM, and IFFT are prior art, and have been described in the present invention related art.
And 2.2, setting a training sequence { T (N) } and a harmonic sequence { a (N) } with the length of N and the period of T, and ensuring that the amplitude value of Fast Fourier Transform (FFT) at the periodic frequency point P is 0 in { d (N) + a (N) }. Where P ═ lN/T, Q ═ N/T and N, P, T, Q are integers, l ═ 0,1, …, T-1, and FFT is the prior art, which has been explained in the invention-related art. The data sequence { d (n) } and the reconciliation sequence { a (n) } satisfy the following relationship:
the expression for the reconciliation sequence { a (n) } is as follows:
and 2.3, summing the { d (n) }, the joint training sequence { t (n) }, and the harmonic sequence { a (n) }obtainedin the step 2.1 through the operation of a formula (2.3), and generating a time domain sending sequence { s (n) }. Taking the nth sampling value S of the mth OFDM symbolm(n) expressed as:
wherein, theta is a real number and takes the value as: 0<θ<1；Dm(n) is the nth OFDM symbol of the mth OFDM symbolAnd m and n are positive integers corresponding to discrete quantization sequence values of sampling values.
Preferably, the channel frequency response rough estimation method based on time domain statistical average Least Square (LS) estimation includes, on the basis of the superimposed training sequence, obtaining time domain statistical average for M (M is a positive integer) OFDM symbols in a frame, and then obtaining a channel frequency response rough estimation by using the existing LS channel estimation. Among them, the LS channel is estimated as the prior art, and has been explained in the invention related art. The rough estimation is completed according to the following steps:
step 3.1, adding the time domain transmission sequence { s (n) } obtained in step 2.3 of claim 2 with the length L in sequencecp(Lcp>L, L is integer and represents channel length), and is placed at the head of data block, and can maximally eliminate inter-block interference (IBI) between data blocks, make N:1 parallel-serial conversion and send them into channel, then make 1: N serial-parallel conversion and remove cyclic prefix (the original added length of the head of removed data block is L)cpThe CP is convenient for the receiving end to demodulate and receive), the data sequence { y ] is obtainedm(n) }. The expression is as follows:
wherein the data sequence vector is ym＝[ym(0),ym(1),…,ym(N-1)]T(ii) a The channel impulse response is: h ism＝[hm(0),hm(1),…,hm(L-1)]T(ii) a t and DmCorresponding to the equivalent cyclic convolution matrix obtained by the training sequence { t (n) } and the data sequence { D)m(n) the resulting equivalent cyclic convolution matrix; the noise column vector is omegam＝[ωm(0),ωm(1),…,ωm(N-1)]TAnd superscript T denotes matrix transpose and denotes product operation. N-dimensional equivalent cyclic convolution matrices t and DmThe matrix representations of (a) are respectively as follows:
step 3.2 for { y obtained in step 3.1m(N), obtaining a receiving end sequence { Y ] through N-point FFTm(k) }; wherein, Ym(k) The frequency domain representation of the nth subcarrier signal of the mth OFDM symbol is as follows:
and Y ism(k)，Hm，WmAre each ym(n)、hm、ωmN-point FFT. Let diag 2]A diagonal matrix is represented and the internal elements are diagonal data. Then T ═ diag [ T (0), T (1), …, T (N-1)]，D＝diag[Dm(0),Dm(1),…,Dm(N-1)]. In addition, the elements T (k) of the matrix T are expressed as follows:
in the formula (3.5), FFT {. cndot } represents an operator of N-point FFT; exp {. } represents the exponential operator of the natural base e;in units of imaginary numbers.
Step 3.3. for the received data sequence { y ] after cyclic prefix removal and before N-point FFT transformationm(n), taking an OFDM symbol with the length of M (the channel impact response h is unchanged in the symbol range), and calculating the time domain average of the taken OFDM symbol by adopting a time domain statistical averaging method to obtain an averaged signal; the time domain expression is:
step 3.4, the LS channel estimation is adopted to obtain the rough estimation of the channel frequency responseWhen M is infinite, { Dm(n) } the time domain mean is 0, and the noise mean is also 0. Therefore, equation (3.6) can be transformed into:
pair type (3.7) equal sign two-side simultaneous left multiplicationObtaining a coarse estimate of the channel impulse responseAndwhere the subscript "LS" denotes the LS channel estimate,the expression of (a) is as follows:
wherein,is composed ofN point F ofFT transform, superscript "-1" indicates the matrix inversion operation.
Preferably, the transform domain quadratic estimation method combining weighted threshold denoising and equalization decision is realized by adopting the following steps:
step 4.1. rough estimation obtained in step 3.4After amplitude phase compensation, windowing function and M-point Discrete Cosine Transform (DCT) processing, obtainingWhere the subscript "c" denotes in the discrete cosine transform domain, the same is indicated below; amplitude phase compensation is performed by multiplying the coarse estimate by a gain factor delta1Completing the process; the windowing function used may be a window function of width M. And a gain factor delta1And the selected window function can adopt a sine window function SIN, and the expressions are respectively:
step 4.2. subjecting the product obtained in step 4.1 toObtaining N point sequence after threshold denoising and zero filling expansionThe time domain threshold denoising process is realized by adopting the following substeps:
step 4.2.1 sequence is concentrated in low frequency band due to signal energyBy means of a low-pass filter (cut-off frequency P)c＝Lcp-1) obtaining a filtered sequence after filtering out the high frequency components
Step 4.2.2 the product of step 3.4Flat-top sampling is adopted, and the sampling period is Ts＝T/(N+Lcp) To obtain a sample sequence gc(i) And (4) solving an amplitude mode of the channel impulse response corresponding to each sampling point. If a sampling sequence has a sampling points with the same amplitude value, the corresponding weight value of the amplitude value is a. In this case, the threshold λ can be obtained by equation (4.3):
wherein, gc(i) Is the channel impulse response corresponding to the ith sampling point, i is an integer, and i is 1,2, …, N; a is1+a2+···+aq＝N。
Step 4.2.3 post-filter sequenceAnd (4) judging according to the formula (4.4), and keeping or setting the sampling point to zero:
step 4.2.4 sequence { G ] from step 4.2.3c(m), zero padding extension to a sequence of N pointsAnd zero paddingThe extension process is in the sequence Gc(M) adding N-M zeros at the end;
step 4.3. the result of step 4.2.4Respectively processed by N-point IDCT transformation, de-windowing (dividing by sine window function SIN) and secondary amplitude phase compensation (multiplying by gain factor delta)2) To obtainWherein the gain factor delta2The expression is as follows:
step 4.4. combining the sequence { Y ] obtained in step 2.4m(k) Obtained in step 4.3Equalizing decision and equalizing resultRepeating the M-point DCT, the low-pass filtering, the weighted threshold denoising, the zero filling expansion and the N-point IDCT to complete the secondary estimation process of the transform domain to obtain the estimation result of the channel frequency responseThe equalization decision is completed according to the following sub-steps:
step 4.4.1. set the equalization resultThe channel frequency response prediction value is obtained, and the frequency domain estimation value of the sending signal is obtained after the zero forcing equalization of the channel frequency response prediction value of the receiving signalThe zero forcing equalization is the prior art, and has been explained in the invention related technology.The expression is as follows:
step 4.4.2, the frequency domain estimated value of the equalized transmission signalAfter data judgment, the data is mapped to the nearest point of QAM constellation diagram to obtain the judgment value of the transmitted signalObtaining frequency channel response decision values simultaneouslyWherein,the expression is as follows:
the decision is based on the following:
1) when the decision valueThen the decision result is correct, i.e. the channel frequency response decision valueFor the actual frequency response value H, X of the channelm(k) N-point FFT data modulated to the kth subcarrier in the mth OFDM symbol;
2) when the decision valueThen the value is decidedA decision error delta exists between the predicted value and the actual frequency response of the channel, and the predicted value of the frequency response of the channel can be corrected through the feedback of the decision error deltaThe channel actual response H is approximated step by step. Wherein, the decision error Δ expression is:
and the decision feedback coefficient ξ is a predicted value of the channel frequency responseAnd ξ is a function of the decision error delta, i.e.Wherein,
step 4.4.3. Joint first DCT estimationDecision valueAnd the decision feedback coefficient ξ is weighted and summed to obtain the channel frequency response equalization resultWherein,the expression is as follows:
where ". x" denotes a product operation. Q1、Q2、Q3Are respectively asξ, all values are in the interval [0,1 ]]Real number of, and Q1+Q2+Q3＝1。
The prior art to which the present invention relates is as follows:
fast Fourier transform/inverse fast Fourier transform (FFT/IFFT) technology, Turbo code coding, pseudorandom interleaving, multi-system quadrature amplitude modulation (M-QAM), a Least Square (LS) channel estimation method, a discrete cosine transform/inverse discrete cosine transform (DCT/IDCT) -based transform domain channel estimation method and a zero forcing equalization technology. The principles of the prior art are described below:
FFT/IFFT technique
The FFT/IFFT technology is the key for realizing modulation and demodulation by the OFDM technology, the FFT/IFFT technology and the OFDM technology are inverse operations with each other, and the low-complexity realization of discrete Fourier transform/inverse discrete Fourier transform (DFT/IDFT) is realized. The modulation and demodulation technique of OFDM can be accomplished by FFT/IFFT technique. An OFDM symbol contains a plurality of modulated subcarriers, which can be expressed as the sum of the plurality of subcarriers, that is:
wherein N is the number of subcarriers; t represents an OFDM symbol duration; diIs a data symbol allocated to each subchannel; f. ofiThe carrier frequency of the ith subcarrier; rect (T) is a rectangular function, and rect (T) is 1, -T/2 ≦ T ≦ T/2; and "" is a product operation, exp {. DEG } represents an exponential operator of a natural base e，The same is expressed below in imaginary units. In formula (1), tsWhen 0 and rect (T) is 1, the signal s (T) is sampled at the rate T/N, which is obtained by T kT/N (k 0,1, …, N-1):
from formula (3), sk(t) is equivalent to the pair diThe IDFT operation of (1) is performed on the receiving end pair sk(t) recovering d by DFT operationiNamely:
therefore, the modulation and demodulation of OFDM can be realized by FFT/IFFT technology, and FFT/IFFT is a fast algorithm of DFT/IDFT.
Turbo code encoding
The Turbo code encoder consists of a component encoder, an interleaver, a puncturing matrix and a multiplexer. The best choice of component codes is the Recursive Systematic Convolutional (RSC) code. Typically both component codes use the same generator matrix. When encoding, the input information sequences of the two component codes are the same, and the input information sequence with the length of N { u }kThe 1 st component encoder is used as the system output while being sent to encodeDirectly to the multiplexer, while { u }kInterlace sequence after interleaver pi { unIt is sent to the 2 nd component encoder. Wherein N is pi (k), N is not less than 0, and k is not more than N-1. Pi (·) is the interleaving mapping function, and N is the interleaving length, i.e., the length of the input information sequence. The input sequences of the two component encoders are only different in code element sequence, and the output check sequences are respectivelyAndin order to improve the code rate and the system spectrum efficiency, two check sequences are obtained after puncturingFinally, willAnd system outputTogether forming a sequence of codewords ck}。
The Turbo coding schematic block diagram is shown in fig. 8.
Pseudo-random interleaving
The implementation steps of the pseudo-random interleaving with the interleaving length of N are as follows: first, an integer i is randomly selected from the set S ═ {1,2, …, N }, and1is correspondingly selected to i1Probability P (i)1) 1/N, i to be selected1Is recorded as pi (1) and simultaneously i is1Deleting the data from the set S to obtain a new set S1(ii) a Secondly, in the k-th step, from the set Sk-1I ≠ i1,i2,…,iN-k+1Randomly select one ikTheir corresponding probability of selection P (i)k) 1/(N-k +1), i to be selectedkIs recorded as pi (k) and simultaneously i iskFrom the set Sk-1Deleting to obtain a new set, and marking as Sk(ii) a Finally, when k is equal to N, pi (N) is obtained, and the corresponding selection probability is P (i)N)＝1，SNFor the empty set, the interleaving process ends.
Multilevel quadrature amplitude modulation (M-QAM)
Quadrature Amplitude Modulation (QAM) is a vector modulation that uses two independent baseband signals for two phasesThe mutually orthogonal co-frequency carriers carry out carrier double-sideband amplitude modulation inhibition, and the modulated signals with frequency spectrum orthogonality in the same broadband are utilized to realize the transmission of two paths of parallel digital information. The modulation and demodulation principle of M-QAM is as follows: the transmitting end converts the information rate into R through serial-parallel conversionbIs divided into two rates RbBinary signal of/2, 2/L level conversion will have two rates of RbBinary signal of/2 becomes rate Rb/[2·lb(L)]Multiplying the level signals by two orthogonal carriers respectively, and adding and summing to obtain M-QAM signals; and at the receiving end, an orthogonal coherent demodulation method is adopted, the received signals are divided into two paths to enter two orthogonal carrier coherent demodulators, the two paths enter a decision device respectively to form L-system signals and output binary signals, and finally, the L-system signals and the binary signals are subjected to parallel-serial conversion to obtain baseband signals. Where lb (·) represents the base-2 logarithmic operator. The M-QAM modulation and demodulation diagram is shown in fig. 9, and "LPF" in the diagram represents a low pass filter.
LS channel estimation method
The criterion of the LS channel estimation method is to minimize the value of the cost function J without considering the influence of noise, where the cost function J is defined as:
J＝(Y-XFh)H(Y-XFh) (4)
wherein, Y ═ Y (0), Y (1), …, Y (N-1)]A vector consisting of output signals demodulated by one OFDM symbol; x ═ diag [ X (0), X (1), …, X (N-1)]A diagonal matrix, diag, formed by a frame of signals output after mapping of the binary input complex sequence x (n)]Representing a diagonal matrix; f is an N-dimensional Fourier transform matrix, and N rows and k columns of elements corresponding to the matrix FThe value ranges of N and k are both [0, N-1 ]]And exp {. } represents an exponential operator of a natural base e,is an imaginary unit; h is the channel impulse response to be estimated,the superscript "H" denotes the conjugate transpose of the matrix.
First, the channel is written in matrix form: y XFh + v;
secondly, to minimize the value of the cost function J, the condition needs to be satisfiedNamely:
finally, the time domain estimation is obtained by simplifying the formula (5)And the frequency response of the channel is obtained from H ═ Fh
Transform domain channel estimation method based on DCT/IDCT
Compared with DFT, DCT, M point data sequence is equivalent to 2M point DFT transform after mirror image expansion, and DCT is a pair of DFT real even functions. Different from DFT, DCT does not generate new high-order components, and the periodic expansion of the sequence is continuous at the periodic edge, meanwhile, DCT has the characteristic of energy concentration, and the performance and the realization complexity are superior to those of DFT. The DCT transform domain channel estimation step is shown in fig. 10.
(1) Data sequence Y received for pilot positionp(k) Obtaining an estimate of the channel frequency response at the pilot subcarriers via LS channel estimation
(2) To pairPerforming M-point DCT to obtainThe expression is as follows:
(3) DCT intra-domain pairZero padding is expanded into an N point sequence to obtainThe expression is as follows:
(4) to pairPerforming N-point IDCT transformation to obtainThe expression is as follows:
zero forcing equalization technique
Channel equalization techniques can be divided into two categories, linear equalization and nonlinear equalization. The linear equalization is suitable for the situations that the channel frequency response characteristic is flat and the intersymbol interference is not serious. The linear equalizer may be implemented by a transversal filter as shown in fig. 11.
To achieve channel equalization, it is critical to calculate the tap coefficients of the transversal filter. The zero forcing equalization is to adjust the tap coefficient of the equalizing filter according to the channel characteristics, so that the total characteristics of the equalizer and the channel are approximate to the ideal channel condition, and the frequency domain shows that the output response only has a value at the central point, thereby eliminating the influence of intersymbol interference. In an OFDM system with intersymbol interference, a matrix representation of the signal transmission process: Y-HX + V, where X, Y, V denotes the frequency domain form of the transmit sequence, the receive sequence, and the additive white gaussian noise, respectively, and H is the frequency domain form of the channel impulse response. The basic idea of zero forcing equalization is to find the least-squares solution of the equation set Y ═ HX, i.e. when | | Y-HX | | luminance2And when a minimum value is taken, solving for X.
Order toNamely, it is
Obtaining by solution:the equalization coefficient matrix of the zero forcing equalizer is obtained by the following steps:
wherein, the superscript H is matrix conjugate transpose, and the superscript-1 is matrix inversion operation.
In the invention, a transmitting sequence is generated by combining a superimposed training sequence and a harmonic sequence, the rough estimation of channel frequency response is obtained by adopting the LS channel estimation method of the prior superimposed training sequence, a noise threshold (the threshold is the weighted average value of the channel response amplitude corresponding to a sampling point) and de-noising processing are set, the transform domain secondary estimation of the channel frequency response is simultaneously carried out by adopting windowed DCT/IDCT interpolation, and an equilibrium judgment method is introduced in the secondary estimation process. The transform domain channel estimation method combining the superimposed training sequence and the time domain denoising eliminates the influence of the data sequence and the training sequence on the channel estimation performance, improves the performance of the existing LS channel estimation, has lower realization complexity, and can be well applied to multi-carrier communication.
Drawings
Fig. 1 is a general framework diagram of the implementation principle of the present invention.
Fig. 2 is a schematic diagram of a generation flow of a transmission sequence { s (n) } and a frame structure of the transmission sequence { s (n) } in the embodiment of the present invention.
Fig. 3 is a structural diagram of a power line carrier multipath channel frequency response rough estimation obtained by using a conventional method according to an embodiment of the present invention.
Fig. 4 is a schematic block diagram of weighted temporal threshold setting according to an embodiment of the present invention.
FIG. 5 is a schematic diagram of the transform domain quadratic estimation process of channel frequency response using windowed DCT/IDCT interpolation according to an embodiment of the present invention.
FIG. 6 shows an embodiment of the invention in which the data sequence Y is jointly receivedm(k) And first DCT domain estimationObtaining the channel frequency response equilibrium valueSchematic representation.
FIG. 7 is a flow chart illustrating an embodiment of the present invention.
FIG. 8 is a block diagram of Turbo coding principles.
Fig. 9 is a schematic diagram of M-QAM modulation and demodulation.
Fig. 10 is a diagram of DCT transform domain channel estimation.
Fig. 11 is a schematic diagram of a linear equalizer implemented by a transversal filter.
Detailed Description
The present invention will be described in further detail below with reference to preferred embodiments and with reference to the attached drawings.
The transform domain secondary estimation method of combined weighted threshold denoising and equalization decision provided by the invention can be applied to a typical power line multi-carrier wired communication or multi-carrier wireless communication system, can quickly and accurately complete the functions of channel estimation and signal detection when clutter interference is severe, and is not limited to the field detailed in the following embodiments. The following describes a specific embodiment of the present invention in detail by selecting a transform domain quadratic estimation method of joint weighted threshold denoising and equalization decision of a typical power line multi-carrier communication system.
The superior embodiment of the invention is realized by the following main steps in sequence:
jointly superposing a training sequence and a harmonic sequence to generate a sending sequence; a typical multi-path channel for power line multi-carrier communication is combined in an OFDM system and the OFDM channel estimation [ J ] of Chirp training sequence is superposed through the existing modified LS channel estimation method (see the methods of Liuqige, Mudaming, Luxianhui and Luxianhui)]Computer engineering and applications, 2011,47(31): 97-100.), yielding a coarse estimate of the powerline channel frequency responseFrom the resulting rough estimateCompensating for its amplitude and phase (with a gain factor delta)1Multiply)Windowing (using a sine window function of width M), M-point DCT transformation and low-pass filtering (cut-off frequency P)c＝Lcp-1) processing to obtain a filtered sequenceAnd sets a cyclic prefix (CP, L)cpLength of cyclic prefix) as a threshold, performing threshold judgment on the result after filtering processing to finish first denoising processing, if the channel response amplitude value corresponding to the sampling point after filtering is greater than a set threshold lambda, reserving the sampling point, otherwise, setting the sampling point to zero; the result after the first denoising processing is Gc(m) } spreading zero padding to N-point sequences(sequence { G)cAdding N-M0 to tail part, and performing N-point IDCT transformation, windowing (removing sine window with width of M), and secondary amplitude phase compensation (and gain factor delta)2Multiplication) to obtain a first transform domain estimateIn combination with Ym(k) (frequency domain signal corresponding to nth subcarrier signal of mth OFDM symbol) andthe equalization decision is obtainedTo pairAnd repeating the M-point DCT transformation, the low-pass filtering, the weighted threshold denoising, the interpolation zero-filling and the N-point IDCT transformation to finally obtain the channel frequency response of the power line and complete the secondary estimation of the transform domain.
The invention eliminates the influence of the data sequence on the channel estimation by the transformation of the data sequence, the rough estimation of the channel frequency response, the combined weighted threshold denoising and the balance judgment, sets an improved threshold with better realization effect and introduces a balance judgment method to complete the transform domain secondary estimation, further eliminates the noise interference, achieves the improvement of the estimation performance and realizes the transform domain secondary estimation method of the combined weighted threshold denoising and the balance judgment with moderate complexity.
The invention provides a transform domain quadratic estimation method combining weighted threshold denoising and equalization decision, which is applied to the embodiment of typical power line multi-carrier communication, and the specific implementation mode can be illustrated by the following legend in sequence.
Fig. 1 is a general block diagram illustrating the principle of the present invention. Wherein, the diagram (a) combines the training sequence and the harmonic sequence to generate the transmission sequence, and adopts the existing LS channel estimation method (the method is shown in Liuqige, Mudamine, Luxian, and Chirp training sequence superposed OFDM channel estimation [ J ] of Chirp training sequence]Computer engineering and applications, 2011,47(31):97-100. "), obtaining a coarse estimate of the channel frequency response; graph (b) is a rough estimation of the channel frequency responseThe process schematic diagram of DCT/IDCT transform domain secondary estimation is completed through DCT/IDCT transform domain interpolation, time domain weighted threshold denoising and equalization decision processing; the diagram (c) depicts a schematic diagram of the link connection between the diagram (a) and the diagram (b), and the represented link relation is as follows: 1) combining the training sequence and the harmonic sequence to generate a sending sequence; 2) sending the sending sequence to the channel, and obtaining the rough estimation of the channel frequency response according to the LS estimation criterion by adopting the time domain statistical averaging method of the superimposed training sequence at the receiving end3) Will roughly estimatePerforming DCT/IDCT interpolation and weighted threshold denoising to complete the second stepA transform domain estimation to obtain4) Joint receiver data sequence Ym(k) And the first transform domain estimation resultIs judged through equalizationAnd will beAnd performing DCT/IDCT interpolation and weighted threshold denoising treatment again to complete the second transform domain estimation.
Fig. 2 is a schematic diagram showing a generation flow of a transmission sequence { s (n) } and a frame structure of the transmission sequence { s (n) } in the embodiment of the present invention. In the present invention, as shown in fig. 2(a), in order to eliminate the influence of the data sequence { d (n) } on the channel estimation, a reconciliation sequence is introduced, the data sequence { d (n)) } is modified through the reconciliation of the reconciliation sequence { a (n)) }, and then a transmission sequence { s (n)) }isgenerated in combination with the training sequence { t (n)) }. The expression for the transmission sequence s (n) isWherein "+" represents a product operation; θ is the power of the training sequence { t (n) }, which takes the values: 0<θ<1; the sequence { d (N) } is a data sequence generated by channel coding/interleaving, M-QAM modulation, 1: N serial-parallel conversion and N-point IFFT conversion of a binary input sequence { x (N) }; the sequence { T (N) } is a training sequence of length N, with a period T (T is a positive integer). FIG. 2(b) is a diagram of the frame structure of the transmission sequence { s (N) }, which is a data sequence of length N { d (N) + a (N) } and a training sequence of length N { t (N) }, and adds a training sequence of length L (N) to their headscpCyclic Prefix (CP) of, and LcpIs an integer multiple of T.
In fig. 2(a), the circle containing "+" represents the summation operation,the circle containing "x" represents the product operation; the reconciliation sequence { a (n) } satisfies the following condition:and isIt is used to eliminate the effect of the training sequence { T (N) } and the data sequence { d (N) } on the channel estimation, and Q ═ N/T is an integer.
As shown in FIG. 3, the present invention adopts the existing channel estimation method (see "Liuqige, Mudamine, Luxianqi, Luyanhui. the OFDM channel estimation of superimposed Chirp training sequence [ J]Computer engineering and applications, 2011,47(31):97-100 "), to accomplish a coarse estimation of the channel frequency response. The method comprises the following implementation steps: 1) binary input sequence { x (n) }, a data sequence { d (n) } is obtained after the processing of the front end of an OFDM baseband system, and a sending sequence { s (n) }isgenerated by combining a tone sum sequence { a (n) } and a training sequence { t (n) }; 2) adding a length L to each OFDM symbolcpThe Cyclic Prefix (CP) is sent out after N:1 parallel-serial conversion, and the added length is L after 1: N serial-parallel conversion and removal at the receiving end through a typical power line channelcpAfter CP, the time domain statistical average is calculated in a certain frame OFDM symbol to obtain the rough estimation of the channel impulse response time domainWill be provided withObtaining a coarse estimate of the channel frequency response by N-point FFTAnd the result after CP removal is ym(N) performing N-point FFT to obtain { Y }m(k)}。
Fig. 4 is a schematic diagram illustrating the setting of the noise threshold λ according to the present invention. The noise threshold lambda setting process is completed by the following steps: 1) to pairAdopting flat top sampling with a sampling period of TsT/N, a sample sequence g is obtainedc(i)}，gc(i) The channel impulse response corresponding to the ith sampling point is represented by i ═ 1,2, …, N; 2) sequentially solving channel impulse response amplitude | g corresponding to each sampling pointc(i) Sorting all amplitude values; 3) recording the number of equal amplitude values as weight (if a certain amplitude value has the same number, there is aqA thenqAs a weight), then a will beqMultiplying by its corresponding amplitude value, as shown in fig. 4; 4) all the weights aqAnd summing the products of the amplitude values and multiplying the products by the reciprocal of the weight sum to obtain the noise threshold lambda.
FIG. 5 shows a rough estimation of the channel frequency response according to the present inventionAnd obtaining the estimation of the channel frequency response after DCT/IDCT transform domain interpolation, time domain weighted threshold denoising and equalization decision processing. The transform domain quadratic estimation process depicted in fig. 5 is implemented by the following steps in sequence: first, rough estimation of the channel frequency response obtained in FIG. 2By compensating for the coarse estimate of amplitude phase (with a gain factor delta)1Multiplication) and multiplication by a sine window function SIN (window function length M) as band limitation to obtain a resultSecondly, toPerforming M-point DCT transform and passing a cut-off frequency of Pc(Pc＝Lcp-1，LcpLength of cyclic prefix), filtering out high-frequency component, comparing amplitude value corresponding to its result with weighted time-domain threshold value lambda, when channel response amplitude corresponding to sampling point obtained by flat-top sampling is greater than lambdaIf not, setting the sampling point to zero; thirdly, zero filling and expanding the judged result into an N-point sequenceAnd performing N-point IDCT conversion, windowing processing and secondary amplitude phase compensation (and gain factor delta)2Multiplication) to obtain the frequency response of the channelFinally, the joint receiving end receives the data sequence { Ym(k) And the first DCT/IDCT estimation resultEqualization decision, secondary DCT/IDCT estimationWherein, the sampling period T of flat-top samplings＝T/(N+Lcp) (ii) a The sine window function SIN is expressed as:k is 0,1,2, …, M; the second DCT/IDCT estimation includes the decision resultThe method comprises the steps of M-point DCT transformation, low-pass filtering, secondary noise threshold judgment, zero filling and expansion into an N-point sequence and N-point IDCT transformation.
FIG. 6 shows a joint receiving end data sequence Y according to an embodiment of the present inventionm(k) And first DCT domain estimationObtaining the channel frequency response equilibrium valueSchematic representation. The equalization decision step depicted in fig. 6: first, an equalization decision result is setFor channel frequency response prediction value, the existing zero forcing equalization technology is adopted to combine the data sequence { Y at the receiving endm(k) Get the estimated value of the frequency domain of the sending signalAnd isIs expressed asSecondly, the frequency domain estimated value of the transmitted signalAfter data judgment, the data is mapped to the nearest point of QAM constellation diagram to obtain the judgment value of the transmitted signalAnd channel frequency response decision valueBoth have the relation ofFinally, the first DCT estimation is combinedChannel frequency response decision valueAnd a decision feedback coefficient ξ (ξ isCorrection factor of (d), function of decision error delta) to obtain a channel frequency response equalization resultThe expression isWherein "+" represents a product operation; q1、Q2、Q3Are respectively asξ, all weights are in the interval [0,1 ]]Real number in (1), and Q1+Q2+Q3＝1。
The data judgment basis is as follows: when sending signal decision valueAnd Xm(k) When the N-point FFT data of the kth subcarrier modulated in the mth OFDM symbol are equal, the decision result is correct, that is, the channel frequency response decision valueIs the channel actual frequency response H; when in useAnd Xm(k) When they are not equal, thenHas a decision error delta with H, and can be corrected by delta feedbackH is approached step by step. Wherein the decision errorDecision feedback coefficient ξ is taken to be the squared relation of decision error ΔGradient of (i), i.e.
Fig. 7 is a schematic flow chart of the embodiment of the present invention. Fig. 7 depicts the main steps implemented by the present embodiment:
the first stage, starting channel estimation process;
the second stage, initializing parameters of various types of data;
in the third stage, a training sequence and a harmonic sequence are jointly superposed to generate a sending sequence;
the fourth stage, the transmission sequence is transmitted through a typical power line multipath channel;
the fifth stage, obtain the rough estimation of the channel response;
the sixth stage, front-end processing of transform domain estimation (including windowing of coarse estimation, amplitude phase compensation, M-point DCT transform and low-pass filtering);
a seventh stage, setting a weighted noise threshold;
the eighth stage, compare the signal channel response amplitude of the sampling point with threshold of the noise threshold presumed, judge whether the sampling point keeps;
the ninth stage, the reserved result is expanded into N point sequence, and the N point IDCT conversion, the windowing processing and the secondary amplitude phase compensation are carried out to obtainIn combination with Ym(k)、Carrying out frequency domain equalization judgment and setting weight Q1、Q2、Q3To obtain the equalization decision result
The tenth stage, for the judgment resultRealizing secondary de-noising, repeating M-point DCT transformation, low-pass filtering, weighted threshold judgment, zero filling and expanding to N-point sequence, and then making N-point IDCT transformation on the result to obtain the result of channel frequency response
The dashed line with arrows in fig. 7 is mainly the decision resultAnd realizing the zero filling expansion process after secondary denoising and threshold judgment.
The invention provides a transform domain secondary estimation method combining weighted threshold denoising and equalization judgment, which comprises a method for generating a sending sequence by jointly superposing a training sequence and a harmonic sequence, a method for obtaining channel frequency response coarse estimation by adopting modified Least Square (LS) channel estimation, and a transform domain secondary estimation method for completing coarse estimation by combining weighted threshold denoising and equalization judgment. The invention is completed by the following steps in sequence: jointly superposing the training sequence and the harmonic sequence to generate a sending sequence; obtaining a channel frequency response rough estimation by adopting a modified LS channel estimation method; sequentially carrying out amplitude phase compensation, windowing function, Discrete Cosine Transform (DCT), low-pass filtering and flat-top sampling on the obtained result, taking the weighted average value of the channel response amplitude of the sampling point as a noise threshold, carrying out threshold judgment on the sampling point, and judging whether the sampling point is reserved or not; and (3) repeating M-point DCT (discrete cosine transform), low-pass filtering, weighted threshold denoising, interpolation zero-filling and N-point IDCT (inverse discrete cosine transform) on the equalization result to complete DCT/IDCT transform domain secondary estimation by zero-filling expansion, the conventional Inverse Discrete Cosine Transform (IDCT) and an improved frequency domain equalization judgment method on the result after threshold judgment to obtain the final channel frequency response. By adopting the method, the multi-path channel frequency response estimation of power line multi-carrier, wireless communication and the like can be realized, noise is denoised twice by combining the weighted threshold and the balance judgment, the estimation performance is higher, and the complexity is moderate. Therefore, the method of the invention is suitable for occasions such as channel estimation, clutter interference detection and suppression and the like when the clutter interference of the wired or wireless channel is severe.
Although the embodiments of the present invention have been described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the method, the scope of which is defined in the appended claims and their equivalents. I.e. by varying the power theta of the training sequence, the length L of the cyclic prefix in the method according to the inventioncpWeighted noise threshold lambda, DCT/IDCT transform point number, and equalization decision weight (Q)1、Q2、Q3) The parameters, which still fall within the scope of the process of the present invention, are still protected by the present patent.
Claims (5)
1. The transform domain quadratic estimation method combining weighted threshold denoising and equalization judgment is characterized by comprising the following steps of:
1) combining the training sequence { t (n) } with the reconciliation sequence { a (n)) } to generate a sending sequence { s (n)) }, wherein the reconciliation sequence { a (n)) } is a random periodic sequence with the same period as { t (n)) };
2) in an OFDM baseband system, a coarse estimation of channel frequency response is obtained by an LS channel estimation method of a superimposed training sequence;
3) amplitude phase compensation by coarse estimation of channel frequency responseThen, after a window function with the width of M and M-point DCT transformation are added, the obtained sequence is processed by low-pass filtering, time domain denoising and sequence zero filling expansion, and the expanded N-point sequence is processedPerforming N-point IDCT conversion, windowing processing and secondary amplitude and phase compensation
4) The obtained results are judged in an equalizing wayRepeating the M-point DCT transformation, the low-pass filtering, the time domain denoising, the interpolation zero-filling and the N-point IDCT transformation to obtain the final estimation of the frequency domain response
2. The transform domain quadratic estimation method combining weighted threshold denoising and equalization decision as claimed in claim 1, wherein: step 1) combining and superposing the training sequence and the harmonic sequence to generate a sending sequence:
after a binary input sequence { x (n) } is processed by the front end of an OFDM baseband system, a data sequence { d (n)) } is obtained, and a training sequence { t (n)) } and a harmonic sequence { a (n)) } are combined to generate a sending sequence { s (n)) }.
3. The transform domain quadratic estimation method combining weighted threshold denoising and equalization decision as claimed in claim 2, wherein: the generation of the transmission sequence by jointly superposing the training sequence and the harmonic sequence is realized by the following steps:
step 2.1, set the binary input sequence { x (N) } with length N to be coded/interleaved, modulated, 1: N serial-parallel converted, A1:A2Is represented by A1Way conversion to A2Road, and A1,A2Is a natural number, and N-point fast FourierAfter inverse transformation processing, obtaining a data sequence { d (n) };
step 2.2, setting a training sequence { T (N) } and a harmonic sequence { a (N) } with the length of N and the period of T, and ensuring that the amplitude value of the fast Fourier transform of { d (N) + a (N) } at the periodic frequency point P is 0; wherein P ═ lN/T, Q ═ N/T and N, P, T, Q are integers, l ═ 0,1, …, T-1; the data sequence { d (n) } and the reconciliation sequence { a (n) } satisfy the following relationship:
the expression for the reconciliation sequence { a (n) } is as follows:
step 2.3, summing up the { d (n) }, the joint training sequence { t (n) }, and the harmonic sequence { a (n) }obtainedin the step 2.1 through the operation of a formula (2.3), and generating a time domain sending sequence { s (n) }; taking the nth sampling value S of the mth OFDM symbolm(n) the expression is:
wherein θ is the power of the training sequence { t (n) }, and is a real number, and its value is: 0<θ<1；DmAnd (n) is a discrete quantization sequence value corresponding to the nth sampling value of the mth OFDM symbol, wherein m and n are positive integers.
4. The transform domain quadratic estimation method combining weighted threshold denoising and equalization decision as claimed in claim 3, wherein: the rough estimation is completed according to the following steps:
step 3.1, adding the time domain sending sequence { s (n) } obtained in step 2.3 with the length L in sequencecpThe cyclic prefix and N:1 parallel-to-serial conversion are sent into a channel, and then the data sequence { y is obtained after 1: N serial-to-parallel conversion and cyclic prefix removalm(n) }, the expression is as follows:
wherein the data sequence vector is ym＝[ym(0),ym(1),…,ym(N-1)]T(ii) a The channel impulse response is: h ism＝[hm(0),hm(1),…,hm(L-1)]T(ii) a t and DmCorresponding to the equivalent cyclic convolution matrix obtained by the training sequence { t (n) } and the data sequence { D)m(n) the resulting equivalent cyclic convolution matrix; the noise column vector is omegam＝[ωm(0),ωm(1),…,ωm(N-1)]TAnd the superscript T represents the matrix transposition, and "+" represents the product operation; n-dimensional equivalent cyclic convolution matrices t and DmThe matrix representations of (a) are respectively as follows:
step 3.2 for { y obtained in step 3.1m(N), obtaining a receiving end sequence { Y ] through N-point FFTm(k) }; wherein, Ym(k) For the nth subcarrier signal y of the mth OFDM symbolm(n) in the frequency domain, the expression is as follows:
wherein, Ym(k)，Hm，WmAre each ym(n)、hm、ωmPerforming FFT on the N points; let diag 2]Representing a diagonal matrix with internal elements as diagonal data, T ═ diag [ T (0), T (1), …, T (N-1)]，D＝diag[Dm(0),Dm(1),…,Dm(N-1)](ii) a In addition, the matrix TThe expression of element T (k) is as follows:
in the formula (3.5), FFT {. cndot } represents an operator of N-point FFT; exp {. } represents the exponential operator of the natural base e;is an imaginary unit;
step 3.3. for the received data sequence { y ] after cyclic prefix removal and before N-point FFT transformationm(n), taking an OFDM symbol with a frame length of M, and averaging the taken OFDM symbol by using a time domain statistical averaging method to obtain an averaged signal, where the time domain expression is:
step 3.4, the LS channel estimation is adopted to obtain the rough estimation of the channel frequency responseWhen M is infinite, { Dm(n) } the temporal mean is 0, and the noise mean is also 0, so equation (3.6) is modified as:
pair type (3.7) equal sign two-side simultaneous left multiplicationObtaining a coarse estimate of the channel impulse responseAndwhere the subscript "LS" denotes the LS channel estimate,the expression of (a) is as follows:
wherein,is composed ofThe N-point FFT of (1) is superscripted with "-1" to indicate the matrix inversion operation.
5. The transform domain quadratic estimation method combining weighted threshold denoising and equalization decision as claimed in claim 4, wherein: the steps 3) and 4) are as follows:
step 4.1. rough estimation obtained in step 3.4After amplitude phase compensation, windowing function and M-point discrete cosine transform processing, obtainingWherein the subscript "c" is in the discrete cosine transform domain; amplitude phase compensation is performed by multiplying the coarse estimate by a gain factor delta1Completing the process; the added window function adopts a window function with the width of M; and a gain factor delta1And the selected window function adopts a sine window function SIN, and the expressions are respectively:
step 4.2. subjecting the product obtained in step 4.1 toAfter threshold denoising and zero filling expansion, obtainingThe time domain threshold denoising process is realized by adopting the following substeps:
step 4.2.1 sequence is concentrated in low frequency band due to signal energyFiltering out high frequency component by low pass filter to obtain filtered sequence
Step 4.2.2 the product of step 3.4Flat-top sampling is adopted, and the sampling period is Ts＝T/(N+Lcp) To obtain a sample sequence gc(i) Calculating amplitude modes of channel impact responses corresponding to the sampling points; if a sampling sequence has a sampling points with the same amplitude value, the corresponding weight value of the amplitude value is a; in this case, the threshold λ is obtained by equation (4.3):
wherein, gc(i) Is the channel impulse response corresponding to the ith sampling point, i is an integer, and i is 1,2, …, N; a is1+a2+···+aq＝N；
Step 4.2.3 post-filter sequenceAnd (4) judging according to the formula (4.4), and keeping or setting the sampling point to zero:
step 4.2.4 sequence { G ] from step 4.2.3c(m), zero padding extension to a sequence of N pointsAnd the zero padding extension process is in the sequence Gc(M) adding N-M zeros at the end;
step 4.3. the result of step 4.2.4Respectively carrying out N-point IDCT conversion, windowing processing and secondary amplitude and phase compensation to obtainWherein the quadratic amplitude phase compensation is performed by multiplying the result of the windowing by a gain factor delta2Is completed with a gain factor delta2The expression is as follows:
step 4.4. combining the sequence { Y ] obtained in step 2.4m(k) Obtained in step 4.3Equalizing decision and equalizing resultRepeating the M-point DCT transformation, the low-pass filtering, the weighted threshold denoising, the interpolation zero-filling and the N-point IDCT transformation to complete the secondary estimation process of the transform domain to obtain the estimation result of the channel frequency responseThe equalization decision is completed according to the following sub-steps:
step 4.4.1. set the equalization resultThe channel frequency response prediction value is obtained, and the frequency domain estimation value of the sending signal is obtained after the zero forcing equalization of the channel frequency response prediction value of the receiving signal The expression is as follows:
step 4.4.2, the frequency domain estimated value of the equalized transmission signalAfter data judgment, the data is mapped to the nearest point of QAM constellation diagram to obtain the judgment value of the transmitted signalObtaining frequency channel response decision values simultaneouslyWherein,the expression is as follows:
the decision is based on the following:
1) when the decision valueThen the decision result is correct, i.e. the channel frequency response decision valueFor the actual frequency response value H, X of the channelm(k) N-point FFT data modulated to the kth subcarrier in the mth OFDM symbol;
2) when the decision valueThen the value is decidedA decision error delta exists between the predicted value and the actual frequency response of the channel, and the predicted value of the frequency response of the channel can be corrected through the feedback of the decision error deltaGradually approaching the actual response H of the channel; wherein, the decision error Δ expression is:
and the decision feedback coefficient ξ is a predicted value of the channel frequency responseAnd ξ is a function of the decision error delta, i.e.
Step 4.4.3. Joint first DCT estimationDecision valueAnd the decision feedback coefficient ξ is weighted and summed to obtain the channel frequency response equalization resultWherein,the expression is as follows:
wherein Q is1、Q2、Q3Are respectively asξ, all values are in the interval [0,1 ]]Real number of, and Q1+Q2+Q3＝1。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611031299.2A CN106789764B (en) | 2016-11-18 | 2016-11-18 | Joint Weighted Threshold denoises and the transform domain quadratic estimate method of balanced judgement |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611031299.2A CN106789764B (en) | 2016-11-18 | 2016-11-18 | Joint Weighted Threshold denoises and the transform domain quadratic estimate method of balanced judgement |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106789764A CN106789764A (en) | 2017-05-31 |
| CN106789764B true CN106789764B (en) | 2019-07-16 |
Family
ID=58971481
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611031299.2A Active CN106789764B (en) | 2016-11-18 | 2016-11-18 | Joint Weighted Threshold denoises and the transform domain quadratic estimate method of balanced judgement |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106789764B (en) |
Families Citing this family (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107294616B (en) * | 2017-06-12 | 2021-03-23 | 西北工业大学 | Dual-extension underwater acoustic channel Doppler diversity communication method based on basis extension model |
| CN107612865B (en) * | 2017-10-12 | 2020-06-02 | 国网天津市电力公司电力科学研究院 | Signal noise reduction method applied to low-voltage power line carrier communication |
| CN107911324B (en) * | 2017-11-17 | 2020-08-04 | 西安电子科技大学 | Frequency offset and phase offset estimation method in single-carrier interleaved frequency division multiple access system |
| CN108566351B (en) * | 2018-04-10 | 2020-03-17 | 西安交通大学 | Secondary phase coupling multi-user frequency difference UNB communication transmission method |
| CN108833311B (en) * | 2018-05-22 | 2021-08-20 | 杭州电子科技大学 | Transform domain quadratic estimation method combining time domain clustering denoising and equalization judgment |
| CN109005139A (en) * | 2018-06-30 | 2018-12-14 | 华南理工大学 | A kind of quadrature amplitude modulation demodulation method and its planisphere optimization method |
| CN109617841B (en) * | 2018-12-14 | 2021-08-10 | 北京遥测技术研究所 | Solid-state power amplifier OFDM signal digital predistortion module in airborne data chain and implementation method |
| US12014710B2 (en) * | 2019-01-14 | 2024-06-18 | Sony Group Corporation | Device, method and computer program for blind source separation and remixing |
| KR102067938B1 (en) * | 2019-01-14 | 2020-01-17 | 박천수 | Zero-Force Equalized Vector Synthesis Absolute Encoder Method and its apparatus |
| CN109905159B (en) * | 2019-01-14 | 2021-03-30 | 电子科技大学 | Multi-channel signal merging method using entropy as weighting |
| CN110071890B (en) * | 2019-04-24 | 2021-11-02 | 哈尔滨工业大学（深圳） | Low peak-to-average power ratio FBMC-OQAM signal processing method and system |
| CN110401611B (en) * | 2019-06-29 | 2021-12-07 | 西南电子技术研究所（中国电子科技集团公司第十研究所） | Method for rapidly detecting CPFSK signal |
| CN110943946B (en) * | 2019-12-04 | 2022-09-13 | 成都国卫通信技术有限公司 | Channel equalization processing system and method |
| CN111371709B (en) * | 2020-03-10 | 2021-09-03 | 西安电子科技大学 | OFDM system channel estimation method based on time domain threshold filtering |
| CN112000918B (en) * | 2020-07-27 | 2023-04-18 | 北京理工大学 | FFT-based frequency estimation interpolation compensation method |
| CN112350746B (en) * | 2020-09-28 | 2021-08-10 | 南京瑞贻电子科技有限公司 | Channel weakening detection system and method for power line carrier communication system |
| CN113595945B (en) * | 2021-08-04 | 2023-06-30 | 重庆邮电大学 | Channel estimation method suitable for PDSCH of 5G system |
| CN114095319B (en) * | 2021-11-05 | 2023-08-18 | 成都中科微信息技术研究院有限公司 | Time-frequency smoothing method for 5G channel estimation by adopting amplitude-phase separation technology |
| CN114844582A (en) * | 2022-04-07 | 2022-08-02 | 展讯通信（上海）有限公司 | Method and device for estimating signal-to-noise ratio of transform domain and communication equipment |
| CN115118560B (en) * | 2022-06-20 | 2023-12-29 | 中国电子科技集团公司第五十四研究所 | Single carrier frequency domain equalization method based on time domain adaptive threshold noise reduction |
| CN116016049B (en) * | 2022-12-27 | 2024-04-19 | 电子科技大学 | Frequency offset estimation method suitable for high dynamic low signal to noise ratio |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102760444A (en) * | 2012-04-25 | 2012-10-31 | 清华大学 | Support vector machine based classification method of base-band time-domain voice-frequency signal |
| CN103152295A (en) * | 2013-03-11 | 2013-06-12 | 浙江万里学院 | Perfect binary sequence pair-based channel estimation and equalization method and system |
| CN103152312A (en) * | 2013-03-26 | 2013-06-12 | 重庆邮电大学 | Power-adjustable partially-superimposed training sequence-based time synchronization system and method for optical orthogonal frequency division multiplexing (OFDM) system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060233289A1 (en) * | 2004-05-21 | 2006-10-19 | Axiocom Inc. | Mixed direct-indirect adaptation procedure applied to receiver filter |
-
2016
- 2016-11-18 CN CN201611031299.2A patent/CN106789764B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102760444A (en) * | 2012-04-25 | 2012-10-31 | 清华大学 | Support vector machine based classification method of base-band time-domain voice-frequency signal |
| CN103152295A (en) * | 2013-03-11 | 2013-06-12 | 浙江万里学院 | Perfect binary sequence pair-based channel estimation and equalization method and system |
| CN103152312A (en) * | 2013-03-26 | 2013-06-12 | 重庆邮电大学 | Power-adjustable partially-superimposed training sequence-based time synchronization system and method for optical orthogonal frequency division multiplexing (OFDM) system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106789764A (en) | 2017-05-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106789764B (en) | Joint Weighted Threshold denoises and the transform domain quadratic estimate method of balanced judgement | |
| CN108833311B (en) | Transform domain quadratic estimation method combining time domain clustering denoising and equalization judgment | |
| EP2605463B1 (en) | Propagation path estimation method and apparatus | |
| JP4099191B2 (en) | Channel estimation apparatus and channel estimation method | |
| CN101621327B (en) | Radio communication method and device in single-carrier transmission system | |
| JP2009532957A (en) | Channel estimation for rapidly dispersive fading channels | |
| JP4147193B2 (en) | Receiving multicarrier spread spectrum signals | |
| JP2006262039A (en) | Propagation path estimation method and propagation path estimation apparatus | |
| Han et al. | Iterative per-vector equalization for orthogonal signal-division multiplexing over time-varying underwater acoustic channels | |
| CN107483373B (en) | Anti-multipath iterative weighting LMMSE channel estimation method and device | |
| JP5428788B2 (en) | Receiving device, receiving method, and receiving program | |
| JP4906721B2 (en) | Digital signal transmission method, digital signal reception method, transmitter and receiver | |
| CN112491756A (en) | RLS channel estimation method for joint channel equalization in FBMC system | |
| CN112003804B (en) | Multipath multivariate differential chaotic shift keying iterative receiving method | |
| CN101322365B (en) | Noise power interpolation in a multi-carrier system | |
| CN113726697A (en) | OTFS symbol detection method based on confidence space dynamic decision | |
| Mannan et al. | Adaptive processing of image using DWT and FFT OFDM in AWGN and Rayleigh channel | |
| EP1757051B1 (en) | Apparatus and method for iteratively estimating a channel transfer function | |
| CN111245589B (en) | Pilot frequency superposition channel estimation method | |
| Ogundile et al. | Improved reliability information for OFDM systems on time-varying frequency-selective fading channels | |
| Chang et al. | Cancellation of ICI by Doppler effect in OFDM systems | |
| Anupriya et al. | Channel estimation for MIMO-OFDM systems using superimposed training In VLSI | |
| JP6491439B2 (en) | Receiving apparatus and program | |
| TWI502936B (en) | Method for signal precoding | |
| Ahmadi et al. | Amplitude reconstruction of clipped OFDM by using DFT-based least squares |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240104 Address after: Room 1502, 15th Floor, Building 1, No. 299 Miaohouwang Road, Xixing Street, Binjiang District, Hangzhou City, Zhejiang Province, 310000 Patentee after: Hangzhou Qilin Technology Co.,Ltd. Address before: No. 1, 2nd Street, Baiyang Street, Jiji Technological Development Zone, Hangzhou City, Zhejiang Province, 310018 Patentee before: HANGZHOU DIANZI University |